India's Banking Sector Releases AI Regulation Draft... High-Risk Models Require Board Approval

The Reserve Bank of India (RBI) released a draft requiring banks and regulated financial companies to implement risk management frameworks for AI and machine learning models. The draft includes kill switches to immediately halt harmful or erroneous AI systems, board-level accountability structures, and a 7月24 deadline for public comments. All models used in banking operations—from simple spreadsheet calculators to complex generative AI—are subject to the requirements. Institutions must maintain lists of all operational models, conduct regular risk assessments, and implement corrective actions if risks exceed thresholds. High-risk models require approval from the board's risk management committee, while all AI systems must include bypass, shutdown, and deactivation features. The RBI also highlighted automation bias as a key risk when employees blindly adopt AI outputs without independent judgment.